On 6 August 2023, an M_S5.5 earthquake struck Pingyuan County, Shandong Province, China. This event represents the largest earthquake in Shandong Province since the M_S5.9 Heze earthquake on 7 November 1983. Investigating its seismogenic structure and aftershock sequence is therefore of significant importance. Large earthquakes are commonly accompanied by abundant aftershocks, and the rapid detection and accurate location of these events are essential for understanding seismogenic processes and assessing seismic hazards.

In recent years, with the rapid growth of seismic observational data, artificial intelli-gence–based aftershock detection methods have been widely adopted to improve detection efficiency. In particular, deep learning–based phase picking approaches have demonstrated clear advantages over traditional techniques. Continuous waveform data recorded by 98 seismic stations between 1 and 11 August 2023 were collected. Phase picking was performed using neural network models, followed by an evaluation of picking accuracy, which confirmed that the results met the requirements for subsequent earthquake location. Using both preliminary and refined relocation procedures, an initial earthquake catalog containing 276 events was obtained.

Phase identification–based detection methods can identify the majority of earthquake events. However, to further improve catalog completeness, previous studies have commonly employed template matching techniques to detect small-magnitude events with low signal-to-noise ratios or those obscured by coda-wave interference. Accordingly, the initial catalog was merged with the unified catalog of the China Earthquake Networks Center, yielding a combined dataset of 290 events. Using this unified catalog as templates, template matching was applied to further expand the event set, ultimately increasing the number of identified earthquakes to 396, all of which were classified as aftershocks of the Pingyuan earthquake. The number of detected aftershocks is approximately 3.2 times that of the manually compiled catalog for the same period. Meanwhile, the magnitude of completeness was reduced from M_L1.8 to M_L1.5, resulting in improved conformity with the Gutenberg–Richter relation and a more reliable estimate of the b value.

The source region of the Pingyuan earthquake is located within the Linqing Depression, where the sedimentary cover is relatively thick and seismic velocities are significantly lower than those in the surrounding uplifted regions. As a result, the shallow velocity structure is highly heterogeneous, making the selection of an appropriate velocity model critical. Based on travel-time inversion, a one-dimensional velocity model was constructed for the source region to better characterize the shallow low-velocity structure. The mainshock was located at a depth of 10.8 km, which is consistent with the typical focal depth range of earthquakes in the North China region.

The relocated aftershocks are predominantly distributed along a north-northeast （NNE） trend, with focal depths mainly concentrated between 8 and 12 km. The mainshock did not occur on any previously mapped faults. Most nearby faults strike southwest–northeast, whereas only a few exhibit east–west orientations. Based on the spatial distribution of the aftershocks and their consistency with regional fault orientations, we infer that the Pingyuan earthquake likely ruptured a blind fault segment located between the Lingxian Fault and the Guanxian fault, intersecting the Linnan Fault. In addition, a small-scale subparallel fault in the vicinity of the mainshock was also activated, further highlighting the structural complexity and multi-fault interactions in this region.